Assessing Substance Use Disorder Symptoms with a Checklist among Primary Care Patients with Opioid Use Disorder and/or Long-Term Opioid Treatment: An Observational Study

Abstract

Background: Primary care (PC) offers an opportunity to treat opioid use disorders (OUD). The Substance Use Symptom Checklist ("Checklist") can assess DSM-5 substance use disorder (SUD) symptoms in PC.

Objective: To test the psychometric properties of the Checklist among PC patients with OUD or long-term opioid therapy (LTOT) in Kaiser Permanente Washington (KPWA).

Design: Observational study using item response theory (IRT) and differential item functioning (DIF) analyses of measurement consistency across age, sex, race and ethnicity, and receipt of treatment.

Patients: Electronic health records (EHR) data were extracted for all adult PC patients visiting KPWA 3/1/15-8/30/2020 who had ≥ 1 Checklist documented and indication of either (a) clinically-recognized OUD (i.e., documented OUD diagnosis and/or OUD medication treatment) or (b) LTOT in the year prior to the checklist.

Main measure: The Checklist includes 11 items reflecting DSM-5 criteria for SUD. We described the prevalence of 2 SUD symptoms reported on the Checklist (consistent with mild-severe DSM-5 SUD). Analyses were conducted in the overall sample and in two subsamples (clinically-recognized OUD and LTOT only).

Key results: Among 2007 eligible patients, 39.9% endorsed ≥ 2 SUD symptoms (74.3% in the clinically-recognized OUD subsample and 13.1% in LTOT subsample). IRT indicated that a unidimensional model for the 11 checklist items had excellent fit (comparative fit index = 0.998) with high item-level discrimination parameters for the overall sample and both subsamples. DIF across age, race and ethnicity, and treatment was observed for one item each, but had minimal impact on expected number of criteria (0-11) patients endorse.

Conclusions: The Substance Use Symptom Checklist measured SUD symptoms consistent with DSM-5 conceptualization (scaled, unidimensional) in patients with clinically-recognized OUD and LTOT and had similar measurement properties across demographic subgroups. The Checklist may support symptom assessment in patients with OUD and diagnosis in patients with LTOT.

Keywords: opioid use disorder; opioids; substance use disorder; symptom assessment.

Figure 1

Item characteristic curves: probability of endorsing each substance use disorder (SUD) symptom by severity of latent SUD among patients with opioid use disorder (OUD) or on long term opioid therapy (LTOT). Each of the 11 SUD criteria is represented as a separate curve. The slope of the curve corresponds to the discrimination parameter a. The point where each line intersects the dashed horizontal line indicates the level of latent SUD severity where patients have a 50% probability of endorsing the criterion, which corresponds to the severity parameter b.

Figure 2

Item characteristic curves: probability of endorsing each substance use disorder (SUD) symptom by severity of latent SUD among subgroups of patients with: opioid use disorder (OUD) and/or OUD treatment (panel 1, n = 875) and LTOT (panel 2, n = 1044). Panels 1 and 2 illustrate item characteristic curves from two separate item response theory models for two different samples. Panels are not directly comparable as they are not on the same scale (different latent severity means and standard deviations). The mean severity of latent SUD in centered at zero in each sample.

Figure 3

Expected number of substance use disorder (SUD) criteria (y-axis) for each demographic or treatment subgroup at the same level of latent SUD severity (x-axis). Each line represents a different demographic or treatment subgroup. Vertical distances between the plotted lines indicate the expected differences in the number of SUD criteria endorsed that are attributable to DIF (note that some lines overlap almost entirely.